Local Agent & Bring-Your-Own-Brain Guide

Solace Buzz is an orchestration harness, not an AI token reseller. You decide where model computation happens: • Local Models (100% Offline): Runs entirely on your hardware. Zero data leaves your machine. • Bring Your Own API Key (Direct Egress): Your desktop app contacts model providers directly. Solace servers never see or hold your keys. • Planned Managed Cloud Brain: Optional managed routing supplied through Solace AGI for teams wanting hands-off operations.

To run free, private, offline models on your laptop or workstation: 1. Install Ollama or run llama-server locally. 2. Download your preferred model: ollama run llama3.2 or ollama run mistral. 3. In Solace Buzz desktop settings, navigate to AI Engines → Local Provider. 4. Set endpoint URL to http://localhost:11434 (or your custom port) and select your model name. 5. Click Test Connection. Solace Buzz validates the local route and displays a green indicator.

If you prefer cloud models (OpenAI, Anthropic, Google Gemini, OpenRouter): 1. Open Settings → AI Engines → Remote Providers. 2. Enter your personal API key. 3. Your key is stored exclusively in your operating system's secure credential keychain (e.g. macOS Keychain, Windows Credential Manager). Important Privacy Guarantee: The Solace central website and hosted relays hold ZERO provider keys. When you execute a task, your desktop client sends the prompt directly to the provider. The central service is completely bypassed.

You can independently verify that your credentials remain secure: • Use network inspection tools (such as Wireshark or browser developer tools) to confirm that API requests flow directly from your machine to the provider's official domain. • Confirm that zero authorization tokens or keys are sent to solacebuzz.com.